Dynamic pricing method for networking charging data fusion platform based on adaptive optimization

Abstract

The invention discloses a networking charging data fusion platform dynamic pricing method based on adaptive optimization. The method comprises the following steps: S1, collecting original data from networking charging equipment and a system; s2, cleaning and preprocessing the collected original data; s3, fusing data from different data sources by adopting a weighted average fusion algorithm; s4, constructing a self-adaptive optimization model, and optimizing a pricing strategy by applying a genetic algorithm; s5, further optimizing the pricing strategy by adopting an improved multi-subswarm particle swarm optimization algorithm, and outputting an optimal pricing strategy; s6, according to the optimized pricing strategy, dynamic pricing is automatically implemented; and S7, monitoring a pricing implementation effect in real time, collecting feedback data, and adjusting the optimization model according to a monitoring result. According to the method, an efficient and scientific optimization scheme can be provided in dynamic pricing of the networking charging platform, and remarkable technical values and economic benefits are brought to practical application.

Description

Technical Field

[0001] The present invention relates to the technical field of networked toll data dynamic pricing, and particularly to a dynamic pricing method for a networked toll data fusion platform based on adaptive optimization. Background Art

[0002] With the continuous development of information technology, especially the application of the Internet of Things, big data, and cloud computing, the networked toll platform has become an important part of fields such as transportation, energy, and public services. The networked toll system mainly realizes the calculation of user fees, payment record management, and real-time monitoring through the connection of devices to the Internet. In order to improve the user experience and platform operation efficiency, many platforms have started to adopt dynamic pricing strategies, dynamically adjusting prices based on user demand, market conditions, and time factors, etc. However, the existing dynamic pricing methods generally have certain limitations and are difficult to meet the complex and changeable market environment and user behavior patterns. There is an urgent need for a more intelligent and flexible pricing mechanism.

[0003] The existing dynamic pricing methods usually rely on rule-based pricing strategies or simple statistical analysis. These traditional methods face obvious defects in practical applications. First of all, the rule-based pricing strategies often cannot flexibly adapt to the rapidly changing market demands. For example, some pricing strategies may only be effective in a certain specific period or for a specific user group, but have poor adaptability to other periods or groups, resulting in a low matching degree between prices and market demands and user behaviors, thereby affecting the platform's revenue and user satisfaction. Secondly, the statistical-based pricing methods usually ignore the personalized needs of users and cannot accurately reflect the consumption behavior characteristics of different users. For example, factors such as different users' payment capabilities, consumption preferences, and behavior habits may have an important impact on the pricing strategy, but traditional methods often fail to fully consider these individual differences, thus limiting the accuracy and effectiveness of the pricing strategy.

[0004] In addition, the existing dynamic pricing methods generally lack the ability of real-time adaptive adjustment. In traditional pricing systems, pricing parameters are usually preset through historical data or empirical rules, lacking an immediate response mechanism to market fluctuations and changes in user behavior. For example, when sudden changes occur in the market (such as a surge in demand, competitors adjusting prices, etc.), the existing pricing strategies often cannot be adjusted in time, resulting in a lag in pricing decisions and being unable to effectively seize market opportunities.

[0005] Meanwhile, traditional methods have weak processing capabilities for large amounts of data and are difficult to handle large-scale information from multiple data sources. Especially in complex networked toll collection systems, how to fuse various types of data in real time and flexibly adjust pricing according to market changes has become a major challenge. To solve the above problems, more and more researchers and enterprises have begun to try to introduce intelligent optimization algorithms to improve existing pricing strategies. For example, intelligent optimization algorithms such as particle swarm optimization and genetic algorithms have been widely used in optimization problems and can find optimal solutions by simulating the evolutionary process in nature. By adaptively adjusting parameters and strategies, these methods can provide more accurate and flexible pricing schemes. However, although intelligent optimization algorithms have strong global search capabilities and self-adaptability, there are still some limitations in practical applications. For example, traditional particle swarm optimization algorithms may have problems such as slow convergence speed or being easily trapped in local optimal solutions when dealing with high-dimensional complex data; while genetic algorithms have large computational overheads and are sensitive to the selection of the initial population, resulting in difficulty in achieving the best effect in practical applications.

[0006] Therefore, when the existing technology solves the dynamic pricing problem of the networked toll collection platform, there are still multiple deficiencies. How to combine advanced adaptive optimization technologies with the actual needs of the networked toll collection system and design an efficient, flexible and real-time responsive pricing mechanism to market changes has become an important topic in current research and practice. By combining the particle swarm optimization algorithm and the genetic algorithm and using adaptive optimization technologies, the limitations of existing methods can be broken through, the accuracy and dynamic adjustment ability of the pricing strategy can be improved, and the overall efficiency and user experience of the platform can be further enhanced. Summary of the Invention

[0007] An object of the present invention is to propose a dynamic pricing method for a networked toll collection data fusion platform based on adaptive optimization. The present invention can provide an efficient and scientific optimization scheme in the dynamic pricing of networked toll collection data, bringing significant technical value and economic benefits to practical applications.

[0008] The dynamic pricing method for a networked toll collection data fusion platform based on adaptive optimization according to an embodiment of the present invention includes the following steps:

[0009] S1. Collect raw data from networked toll collection devices and systems;

[0010] S2. Clean and preprocess the collected raw data;

[0011] S3. Use a weighted average fusion algorithm to fuse data from different data sources;

[0012] S4. Construct an adaptive optimization model, apply the genetic algorithm to optimize the pricing strategy, and automatically adjust the pricing parameters in combination with a fuzzy logic control mechanism;

[0013] S5. Further optimize the pricing strategy by using an improved multi - subgroup particle swarm optimization algorithm. The improved multi - subgroup particle swarm optimization algorithm divides the entire particle swarm into several subgroups, shares information between subgroups, and in each iteration, the subgroups will regularly exchange information of the global optimal particle. Each subgroup is responsible for searching different solution space regions and uses a local optimization strategy to find the optimal solution. When the optimization termination condition is reached, the optimal pricing strategy is output;

[0014] S6. Automatically implement dynamic pricing according to the optimized pricing strategy, and set hierarchical pricing according to different user types, consumption behaviors, and time periods;

[0015] S7. Monitor the implementation effect of pricing in real - time and collect feedback data, adjust and optimize the model according to the monitoring results, and continuously improve the pricing strategy to maximize the platform benefits.

[0016] Optionally, S1 includes the following steps:

[0017] S11. Real - time collect raw data through networked toll collection devices and systems. The raw data includes user behavior data, payment records, traffic data, and time - dimension data;

[0018] S12. Process the user behavior data to extract the user's access frequency, stay time, payment amount, and device information feature data;

[0019] S13. Process the payment record data to extract and analyze the transaction time, transaction amount, payment method, and the user's payment habit feature data;

[0020] S14. Process the traffic data to analyze the network access frequency, data transmission volume, and request type behavior features of the device;

[0021] S15. Process the time - dimension data to analyze the charging behavior, user activity, and payment mode in different time periods, and obtain the traffic change trend in different time periods.

[0022] Optionally, S2 includes the following steps:

[0023] S21. Clean and process the collected raw data to remove noise data;

[0024] S22. Fill in the missing values using the mean filling method;

[0025] S23. Standardize all the collected user behavior data, payment records, traffic data, and time - dimension data;

[0026] S24. For time - series data, fill in the missing time - series data by interpolation method;

[0027] S25. Identify and remove outliers using an anomaly detection algorithm, and verify the processed data.

[0028] Optionally, the S3 includes the following steps:

[0029] S31. Use a weighted average fusion algorithm to fuse data from different data sources. The weighted average fusion algorithm is:

[0030]

[0031] where D f is the fused data, D i is the processed data of the i-th data source, w i is the weight of this data source, and n is the total number of data;

[0032] S32. Align the timestamps between data sources, use the time window technology for synchronization in the time dimension, and match each data source in the time dimension;

[0033] S33. According to the characteristics of time series data, use time series data fusion technology to fuse continuous data;

[0034] S34. Integrate the dimensions of the fused data, and use the principal component analysis method to reduce the dimension of the data.

[0035] Optionally, the S4 includes the following steps:

[0036] S41. Based on an adaptive optimization model, use a genetic algorithm to optimize the pricing strategy. The fitness function of the genetic algorithm is:

[0037]

[0038] where F is the fitness function, P i is the revenue of the i-th pricing strategy, C i is the cost of the i-th pricing strategy, w i is the weight, and n is the total number of strategies;

[0039] S42. In the genetic algorithm, use selection, crossover, and mutation operations to generate new pricing strategies. The selection operation is based on the fitness function value, and strategies with high revenue are preferentially selected for crossover and mutation;

[0040] S43. Combine a fuzzy logic control mechanism to automatically adjust pricing parameters according to real-time data input. The fuzzy control rule is:

[0041] New_Price=Base_Price+ΔP;

[0042] Among them, New_Price is the adjusted pricing, Base_Price is the basic pricing, and ΔP is the pricing adjustment value generated according to the fuzzy logic control mechanism;

[0043] S44. Evaluate the effectiveness of each pricing strategy based on the pricing strategy generated by the genetic algorithm and the fuzzy logic control mechanism.

[0044] Optionally, the S5 includes the following steps:

[0045] S51. Use the improved multi-subgroup particle swarm optimization algorithm to globally optimize the pricing strategy. The multi-subgroup particle swarm optimization algorithm divides the entire particle swarm into several subgroups, and each subgroup searches independently for different solution space regions. Each particle represents a pricing strategy, and the pricing strategy includes the basic pricing, the pricing fluctuation range, and the discount amplitude parameter;

[0046] S52. Define the objective function as:

[0047]

[0048] Among them, F is the objective function value, P i is the revenue of the i-th pricing strategy, C i is the cost of the i-th pricing strategy, w i is the weight, ΔP j is the price fluctuation range of the j-th strategy, γ is the fluctuation penalty coefficient, L is the local search improvement term, κ is the local improvement weight, n is the number of strategies, and m is the number of price fluctuation terms;

[0049] S53. Initialize the position and velocity of each particle in the subgroup. The initial position of each particle is randomly distributed within the predetermined search space, and the initial velocity satisfies:

[0050]

[0051] Among them, U(-V max ,V max ) is a uniform distribution on the interval [-V max ,V max ;

[0052] S54. Update the particle velocity, using the dynamic inertia weight w(k) and the local gradient correction term. The update formula is:

[0053]

[0054] Among them, is the velocity of the i-th particle in the k-th generation, is the position of the $i$-th particle, is the historical best position of the particle, is the best position of the subgroup where it is located, $c$ 1 ,$c$ 2 ,$c$ 3 are learning factors, $r$ 1 ,$r$ 2 are random variables, is the gradient correction term based on local search at the particle position ;

[0055] S55. Adopt a local search strategy for the particles within the subgroup, use the gradient descent method to fine-tune the local optimal solution, calculate the local improvement term $L$ so that the local search obtains a better pricing strategy, and adjust the current particle position:

[0056]

[0057] where, is the position of the $i$-th particle in the $k$-th generation, is the updated particle velocity, is the position of the $i$-th particle in the $(k + 1)$-th generation;

[0058] S56. According to the search results of the particle swarm, adopt the pricing strategy of the global optimal particle as the current optimization strategy. The update formula for the global optimal solution is:

[0059]

[0060] where, is the position of the global optimal particle, is the fitness value of the $i$-th particle in the $(k + 1)$-th generation;

[0061] S57. Dynamically adjust the size and search range of the particle swarm during the optimization process, and make adaptive adjustments according to market feedback and optimization effects. The adjustment formulas for the particle swarm size and search range are:

[0062]

[0063] where, $N$ new is the new particle swarm size, $N$ old is the current particle swarm size, $\beta$ is the adjustment coefficient, $\Delta F$ i is the change in the fitness of the $i$-th particle, represents the sum of the fitness changes of all particles;

[0064] S58. When the fitness value of the global optimal particle reaches the set threshold $\theta$, terminate the optimization process, output the optimal pricing strategy. The judgment formula for the termination condition is:

[0065]

[0066] Among them, Terminate is the termination condition flag. True indicates triggering the stop of the optimization process, and False indicates not triggering the stop of the optimization process.

[0067] Optionally, S6 includes the following steps:

[0068] S61. Automatically implement dynamic pricing according to the optimized pricing strategy. The pricing adjustment formula is:

[0069] New_Price = Base_Price + ΔP + ΔU + ΔT;

[0070] Among them, New_Price is the adjusted price, Base_Price is the basic price, ΔP is the price fluctuation adjustment generated according to market feedback, ΔU is the personalized price adjustment generated according to user consumption behavior, and ΔT is the time period adjustment generated according to the time period;

[0071] S62. Calculate the personalized pricing adjustment value for each user using the weighted average method based on the user's historical behavior data;

[0072] S63. Perform time period adjustment according to user behavior and market feedback. The time period adjustment formula is:

[0073]

[0074] Among them, ΔT is the time period adjustment, α and β are adjustment coefficients, and Demand is the demand volume;

[0075] S64. Apply the dynamic pricing generated by combining the adjustment values to the charging platform.

[0076] Optionally, S7 includes the following steps:

[0077] S71. Monitor the implementation effect of pricing in real time. The monitoring content includes indicators such as platform revenue, user satisfaction, transaction volume, and price fluctuation. The monitoring formula is:

[0078]

[0079] Among them, M is the overall monitoring indicator, m i is the i-th monitoring indicator, w i is the weight of the i-th monitoring indicator, and n is the number of monitoring indicators;

[0080] S72. Process the collected feedback data and calculate the change rate of each monitoring indicator. The change rate formula is:

[0081]

[0082] Among them, Δm i The change rate of the i-th monitoring index is the new period index value is the old period index value;

[0083] S73. According to the change rate of the monitoring data, determine whether the current pricing strategy reaches the expected goal. If the change rate exceeds the set threshold, trigger the adjustment mechanism;

[0084] S74. Set the feedback threshold. If the user satisfaction in the feedback data is lower than the feedback threshold, adjust the pricing strategy according to the market demand fluctuation and user behavior. The adjustment formula is:

[0085] ΔP = α·(Feedback - Threshold);

[0086] Among them, ΔP is the adjusted pricing change amount, α is the adjustment coefficient, Feedback is the user satisfaction, and Threshold is the preset feedback threshold;

[0087] S75. According to the market and user feedback, update the parameters of the optimization model in real time.

[0088] The beneficial effects of the present invention are:

[0089] (1) By introducing an adaptive optimization algorithm, especially the combination of the improved particle swarm optimization algorithm and the genetic algorithm, the present invention effectively overcomes the problems existing in the traditional methods, such as pricing lag, poor adaptability, and insufficient handling of user personalized needs.

[0090] (2) By combining user behavior data with market feedback, the present invention dynamically optimizes the pricing strategy, realizes personalized pricing for different user groups. This personalized pricing not only improves the user experience, but also effectively improves the platform's customer loyalty and user satisfaction. For example, for high-frequency consumption users and low-frequency consumption users, the system can adjust the price according to their historical consumption behavior and payment records, making the pricing more flexible and differentiated. By accurately identifying user needs and adjusting the price, the present invention can help the platform maximize user value and platform benefits.

[0091] (3) The present invention can effectively improve the stability and sustainability of the pricing strategy. Through real-time monitoring and feedback mechanisms, the platform can adjust the pricing strategy in a timely manner according to market fluctuations and user feedback, avoiding revenue losses caused by pricing lag or inflexibility. The improvement of this algorithm not only makes the pricing process more efficient, but also can avoid the problem of falling into local optimal solutions commonly found in traditional optimization methods, thus achieving a global optimal pricing scheme. Description of the Drawings

[0092] The accompanying drawings are used to provide a further understanding of the present invention and form a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention. In the accompanying drawings:

[0093] Figure 1 is a flowchart of the dynamic pricing method for the networked toll collection data fusion platform based on adaptive optimization proposed by the present invention;

[0094] Figure 2 is a flowchart of the improved multi - subgroup particle swarm optimization algorithm in the dynamic pricing method for the networked toll collection data fusion platform based on adaptive optimization proposed by the present invention. Detailed implementation manners

[0095] Now, the present invention will be further described in detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0096] Referring to Figure 1 - Figure 2 , the dynamic pricing method for the networked toll collection data fusion platform based on adaptive optimization includes the following steps:

[0097] S1. Collect raw data from networked toll collection devices and systems;

[0098] S2. Clean and preprocess the collected raw data;

[0099] S3. Use the weighted average fusion algorithm to fuse data from different data sources;

[0100] S4. Build an adaptive optimization model, apply the genetic algorithm to optimize the pricing strategy, and automatically adjust the pricing parameters in combination with the fuzzy logic control mechanism;

[0101] S5. Use the improved multi - subgroup particle swarm optimization algorithm to further optimize the pricing strategy. The improved multi - subgroup particle swarm optimization algorithm divides the entire particle swarm into several subgroups, shares information between subgroups, and in each round of iteration, the subgroups will regularly exchange information of the global optimal particle. Each subgroup is responsible for searching different solution space regions, and uses a local optimization strategy to find the optimal solution. When the optimization termination condition is reached, the optimal pricing strategy is output;

[0102] S6. Automatically implement dynamic pricing according to the optimized pricing strategy, and set hierarchical pricing according to different user types, consumption behaviors, and time periods;

[0103] S7. Real - time monitor the implementation effect of pricing and collect feedback data, adjust the optimization model according to the monitoring results, and continuously improve the pricing strategy to maximize the platform benefits.

[0104] In this embodiment, S1 includes the following steps:

[0105] S11. Real-time collect raw data through the network toll collection device and system, where the raw data includes user behavior data, payment records, traffic data, and time dimension data;

[0106] S12. Process the user behavior data to extract the user's access frequency, stay time, payment amount, and device information feature data;

[0107] S13. Process the payment record data to extract and analyze the transaction time, transaction amount, payment method, and the user's payment habit feature data;

[0108] S14. Process the traffic data to analyze the network access frequency, data transmission volume, and request type behavior characteristics of the device;

[0109] S15. Process the time dimension data to analyze the toll collection behavior, user activity, and payment mode in different time periods, and obtain the traffic change trend in different time periods.

[0110] In this embodiment, S2 includes the following steps:

[0111] S21. Clean the collected raw data to remove noise data;

[0112] S22. Fill in the missing values using the mean filling method;

[0113] S23. Standardize all the collected user behavior data, payment records, traffic data, and time dimension data;

[0114] S24. For time series data, fill in the missing time series data by interpolation method;

[0115] S25. Apply the anomaly detection algorithm to identify and remove outliers, and verify the processed data.

[0116] In this embodiment, S3 includes the following steps:

[0117] S31. Use the weighted average fusion algorithm to fuse the data from different data sources. The weighted average fusion algorithm is:

[0118]

[0119] where D f is the fused data, D i is the processed data of the i-th data source, w i is the weight of this data source, and n is the total number of data;

[0120] S32. Align the timestamps between data sources, use the time window technique for time dimension synchronization, and match each data source in the time dimension;

[0121] S33. According to the characteristics of time series data, adopt time series data fusion technology to fuse continuous data;

[0122] S34. Integrate the dimensions of the fused data and use the principal component analysis method to reduce the data dimensions.

[0123] In this embodiment, S4 includes the following steps:

[0124] S41. Based on the adaptive optimization model, use the genetic algorithm to optimize the pricing strategy. The fitness function of the genetic algorithm is:

[0125]

[0126] where F is the fitness function, P i is the revenue of the i-th pricing strategy, C i is the cost of the i-th pricing strategy, w i is the weight, and n is the total number of strategies;

[0127] S42. In the genetic algorithm, use selection, crossover, and mutation operations to generate new pricing strategies. The selection operation is based on the fitness function value, and strategies with high revenue are preferentially selected for crossover and mutation;

[0128] S43. Combine the fuzzy logic control mechanism and automatically adjust the pricing parameters according to real-time data input. The fuzzy control rule is:

[0129] New_Price = Base_Price + ΔP;

[0130] where New_Price is the adjusted pricing, Base_Price is the base pricing, and ΔP is the pricing adjustment value generated according to the fuzzy logic control mechanism;

[0131] S44. Based on the pricing strategies generated by the genetic algorithm and the fuzzy logic control mechanism, evaluate the effectiveness of each pricing strategy.

[0132] In this embodiment, S5 includes the following steps:

[0133] S51. Use the improved multi-subgroup particle swarm optimization algorithm to globally optimize the pricing strategy. The multi-subgroup particle swarm optimization algorithm divides the entire particle swarm into several subgroups, and each subgroup independently searches in different solution space regions. Each particle represents a pricing strategy, and the pricing strategy includes base pricing, pricing fluctuation range, and discount amplitude parameter;

[0134] S52. Define the objective function as:

[0135]

[0136] where F is the objective function value, P i is the revenue of the i-th pricing strategy, C i is the cost of the i-th pricing strategy, w i is the weight, ΔP j is the price fluctuation range of the j-th strategy, γ is the fluctuation penalty coefficient, L is the local search improvement term, κ is the local improvement weight, n is the number of strategies, and m is the number of price fluctuation terms;

[0137] S53. Initialize the position and velocity of each particle in the subgroup. The initial position of each particle is randomly distributed within a predetermined search space, and the initial velocity satisfies:

[0138]

[0139] where U(-V max , V max ) is a uniform distribution on the interval [-V max , V max ;

[0140] S54. Update the particle velocity. Use the dynamic inertia weight w(k) and the local gradient correction term. The update formula is:

[0141]

[0142] where is the velocity of the i-th particle in the k-th generation, is the position of the i-th particle, is the historical optimal position of the particle, is the optimal position of the subgroup where it is located, c 1 , c 2 , c 3 are the learning factors, r 1 , r 2 are random variables, is the gradient correction term based on local search at the particle position ;

[0143] S55. Adopt a local search strategy for the particles within the subgroup. Use the gradient descent method to fine-tune the local optimal solution, calculate the local improvement term L to obtain a better pricing strategy through local search, and adjust the current particle position:

[0144]

[0145] Among them, is the position of the i-th particle in the k-th generation, is the updated particle velocity, is the position of the i-th particle in the (k + 1)-th generation;

[0146] S56. According to the search results of the particle swarm, adopt the pricing strategy of the global optimal particle as the current optimization strategy. The update formula of the global optimal solution is:

[0147]

[0148] Among them, is the position of the global optimal particle, is the fitness value of the i-th particle in the (k + 1)-th generation;

[0149] S57. Dynamically adjust the size and search range of the particle swarm during the optimization process, and make adaptive adjustments according to market feedback and optimization effects. The adjustment formulas for the particle swarm size and search range are:

[0150]

[0151] Among them, N new is the new particle swarm size, N old is the current particle swarm size, β is the adjustment coefficient, and ΔF i is the fitness change of the i-th particle, represents the sum of the fitness changes of all particles;

[0152] S58. When the fitness value of the global optimal particle reaches the set threshold θ, terminate the optimization process and output the optimal pricing strategy. The judgment formula for the termination condition is:

[0153]

[0154] Among them, Terminate is the termination condition flag, True indicates triggering the stop of the optimization process, and False indicates not triggering the stop of the optimization process.

[0155] In this embodiment, S6 includes the following steps:

[0156] S61. Automatically implement dynamic pricing according to the optimized pricing strategy. The pricing adjustment formula is:

[0157] New_Price = Base_Price + ΔP + ΔU + ΔT;

[0158] Among them, New_Price is the adjusted pricing, Base_Price is the basic pricing, ΔP is the price fluctuation adjustment generated according to market feedback, ΔU is the personalized price adjustment generated according to user consumption behavior, and ΔT is the time period adjustment generated according to the time period;

[0159] S62. Calculate the personalized pricing adjustment value for each user by using the weighted average method based on the user's historical behavior data;

[0160] S63. Conduct time period adjustment according to user behavior and market feedback. The time period adjustment formula is:

[0161]

[0162] Among them, ΔT is the time period adjustment, α and β are adjustment coefficients, and Demand is the demand volume;

[0163] S64. Apply the dynamic pricing generated by combining the adjustment values to the charging platform.

[0164] In this embodiment, S7 includes the following steps:

[0165] S71. Monitor the implementation effect of the pricing in real time. The monitoring content includes indicators such as platform revenue, user satisfaction, transaction volume, and price fluctuation situation. The monitoring formula is:

[0166]

[0167] Among them, M is the overall monitoring indicator, m i is the i-th monitoring indicator, w i is the weight of the i-th monitoring indicator, and n is the number of monitoring indicators;

[0168] S72. Process the collected feedback data and calculate the change rate of each monitoring indicator. The change rate formula is:

[0169]

[0170] Among them, Δm i is the change rate of the i-th monitoring indicator, is the new period indicator value, is the old period indicator value;

[0171] S73. Judge whether the current pricing strategy meets the expected goal according to the change rate of the monitoring data. If the change rate exceeds the set threshold, trigger the adjustment mechanism;

[0172] S74. Set the feedback threshold. If the user satisfaction in the feedback data is lower than the feedback threshold, adjust the pricing strategy according to market demand fluctuations and user behavior. The adjustment formula is:

[0173] ΔP = α·(Feedback - Threshold);

[0174] Where ΔP is the adjusted pricing change, α is the adjustment coefficient, Feedback is the user satisfaction, and Threshold is the preset feedback threshold;

[0175] S75. Update and optimize the parameters of the model in real time according to market and user feedback.

[0176] Example:

[0177] In the shared bicycle platform of a certain city, this platform adopts a dynamic pricing method for an Internet - connected charging data fusion platform based on adaptive optimization. The platform obtains various types of data in real time by accessing Internet of Things devices and cloud data management systems, including users' riding records, payment information, vehicle locations, usage time periods, meteorological data, traffic flow information, etc. The goal of the platform is to formulate a more refined and dynamic pricing strategy based on these data to maximize the platform's revenue, while improving user satisfaction and platform competitiveness.

[0178] Under the traditional pricing strategy, the pricing of the platform is usually fixed. For example, during the peak daytime period, the bicycle rental price may be set at 3 yuan per hour, while it drops to 1 yuan per hour during the low - demand nighttime period. However, with the gradual diversification of the user group and the complexity of the market environment, a single pricing model clearly can no longer meet the ever - changing needs. Especially during some special periods (such as holidays, bad weather, emergencies, etc.), the lag and inflexibility of the traditional pricing method lead to a decline in the platform's profitability and even user loss.

[0179] Therefore, the implementer adopts the method proposed by the present invention. First, the platform collects user behavior data, payment records, traffic data, and time - dimension data through multiple Internet - connected devices (such as smart locks, GPS positioning systems, mobile payment terminals, etc.). For example, whenever a user uses a bicycle, the platform records information such as the user's start time, end time, riding route, payment amount, etc., and also collects local weather data, traffic condition data, etc. The platform cleans and pre - processes the collected data, including removing noise data, filling in missing values, normalizing data formats, etc., to ensure the data quality for analysis and use.

[0180] Next, the platform uses the weighted average fusion algorithm and time-series data fusion technology to fuse multiple data sources to ensure the consistency of all data in the time dimension. For example, the platform will comprehensively evaluate according to real-time traffic flow data, weather data, historical usage data, user types, etc., and combine the weighted average algorithm to determine the influence weights of various data on the pricing strategy. Then, based on the adaptive optimization model, the platform uses an improved genetic algorithm to optimize the pricing strategy. Through the training and simulation of a large amount of historical data, the genetic algorithm can adaptively adjust the pricing parameters, enabling the platform to dynamically adjust the pricing according to real-time market demand, user behavior, environmental changes, etc. For example, during high-demand periods (such as peak commuting hours), the platform can automatically increase the bike rental fee; while during low-demand periods (such as late at night or in bad weather), the pricing is reduced to attract more users to use the bikes.

[0181] In addition, the platform also combines the particle swarm optimization algorithm for further pricing optimization to ensure the stability and adaptability of the pricing model in a dynamic market environment. Through simulation and optimization, the platform can better cope with complex market changes and avoid falling into local optimal solutions. Finally, the platform implements dynamic pricing according to the optimized pricing strategy through real-time monitoring and data feedback mechanisms, and continuously adjusts and optimizes the model to maximize the platform's benefits. When users use the bikes, the system will adjust the rental fee according to real-time data such as current market demand, weather, and time period, and display the current pricing to the users.

[0182] In the actual application process, this method has been verified on the bike-sharing platform in a certain city and achieved remarkable results. The following are the comparison data of the platform before and after implementing the adaptive optimization pricing method:

[0183] Table 1 Comparison of the revenues of the bike-sharing platform using the present invention and the traditional method

[0184] Time period Average daily income before implementation (yuan) Average daily income after implementation (yuan) Income growth rate (%) Daytime on weekdays 3000 4500 50％ Nighttime on weekdays 1200 1300 8.33％ Daytime on weekends 5000 6500 30％ Nighttime on weekends 1500 2000 33.33％ Holidays 8000 10000 25％

[0185] In the whole embodiment, the implementer not only solves the problems of market fluctuations and user demand uncertainty faced by the bike-sharing platform in dynamic pricing through the method of the present invention, but also improves the accuracy and adaptability of the pricing strategy, realizing the efficient operation of the bike-sharing platform and the improvement of user satisfaction.

[0186] In view of the actual operation requirements of the shared bicycle platform, this invention comprehensively considers multi-dimensional data such as user behavior, demand fluctuations, and time factors, and establishes a multi-objective optimization model. By dynamically adjusting the weights of pricing strategies and setting flexible constraints, this invention realizes the optimization of platform revenue management while ensuring user satisfaction. During high-demand periods, the system can dynamically adjust rental fees, and during low-demand periods, it attracts more users to use bicycles by reducing prices, effectively avoiding resource waste and revenue losses caused by excessively high or low prices.

[0187] As mentioned above, the above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

Claims

1. A dynamic pricing method for an online charging data fusion platform based on adaptive optimization, characterized in that: The steps include: S1. Collect raw data from networked toll collection devices and systems; S2, cleaning and preprocessing the collected raw data; S3, using weighted average fusion algorithm to fuse data from different data sources; S4. Build an adaptive optimization model, apply genetic algorithm to optimize pricing strategy, and automatically adjust pricing parameters in combination with fuzzy logic control mechanism; S5. Using an improved multi-subgroup particle swarm optimization algorithm to further optimize the pricing strategy. The improved multi-subgroup particle swarm optimization algorithm divides the entire particle swarm into several subgroups. The subgroups share information. In each round of iteration, the subgroups regularly exchange information about the global optimal particle. Each subgroup is responsible for searching different solution space regions and using a local optimization strategy to find the optimal solution. When the optimization termination condition is reached, the optimal pricing strategy is output. S6. Automatically implement dynamic pricing based on the optimized pricing strategy, and set tiered pricing based on different user types, consumption behaviors and time periods; S7. Monitor the pricing implementation effect in real time and collect feedback data, adjust the optimization model according to the monitoring results, and continuously improve the pricing strategy to maximize the platform benefits.

2. The method for dynamic pricing of online charging data fusion platform based on adaptive optimization according to claim 1 is characterized in that: The S1 comprises the following steps: S11. Collecting raw data in real time through networked charging equipment and systems, wherein the raw data includes user behavior data, payment records, flow data, and time dimension data; S12, processing the user behavior data to extract the user's access frequency, stay time, payment amount, and device information feature data; S13, processing the payment record data, extracting and analyzing transaction time, transaction amount, payment method and user's payment habit characteristic data; S14, processing the traffic data to analyze the network access frequency, data transmission volume, and request type behavior characteristics of the device; S15. Process the time dimension data, analyze charging behaviors, user activity and payment patterns in different time periods, and obtain traffic change trends in different time periods.

3. The method for dynamic pricing of online charging data fusion platform based on adaptive optimization according to claim 1 is characterized in that: The S2 comprises the following steps: S21, cleaning the collected raw data to remove noise data; S22, fill in missing values ​​using mean filling method; S23. Standardize all collected user behavior data, payment records, traffic data, and time dimension data; S24. For time series data, fill in the missing time series data by interpolation; S25. Apply anomaly detection algorithms to identify and remove outliers, and verify the processed data.

4. The method for dynamic pricing of online charging data fusion platform based on adaptive optimization according to claim 1 is characterized in that: The S3 comprises the following steps: S31, using a weighted average fusion algorithm to fuse data from different data sources, the weighted average fusion algorithm is: Among them, D f is the fused data, D i is the processed data of the ith data source, w i is the weight of the data source, and n is the total number of data; S32, aligning the timestamps between the data sources, using the time window technology to synchronize the time dimension, and matching the data sources in the time dimension; S33. According to the characteristics of time series data, the time series data fusion technology is used to fuse the continuous data; S34. Perform dimension integration on the fused data and use principal component analysis to reduce the dimension of the data.

5. The method for dynamic pricing of online charging data fusion platform based on adaptive optimization according to claim 1 is characterized in that: The S4 comprises the following steps: S41. Based on the adaptive optimization model, the genetic algorithm is used to optimize the pricing strategy. The fitness function of the genetic algorithm is: Among them, F is the fitness function, P i is the profit of the i-th pricing strategy, C i is the cost of the i-th pricing strategy, w i is the weight, n is the total number of strategies; S42. In the genetic algorithm, new pricing strategies are generated using selection, crossover and mutation operations. The selection operation is based on the fitness function value, and strategies with high returns are prioritized for crossover and mutation. S43, combined with fuzzy logic control mechanism, automatically adjust pricing parameters according to real-time data input, the fuzzy control rules are: New_Price = Base_Price + ΔP; Among them, New_Price is the adjusted price, Base_Price is the base price, and ΔP is the price adjustment value generated according to the fuzzy logic control mechanism; S44. Based on the pricing strategies generated by genetic algorithm and fuzzy logic control mechanism, the effectiveness of each pricing strategy is evaluated.

6. The method for dynamic pricing of online charging data fusion platform based on adaptive optimization according to claim 1 is characterized in that: The S5 comprises the following steps: S51, using an improved multi-subgroup particle swarm optimization algorithm to globally optimize the pricing strategy, wherein the multi-subgroup particle swarm optimization algorithm divides the entire particle swarm into several subgroups, each subgroup independently searches for different solution space regions, and each particle represents a pricing strategy, wherein the pricing strategy includes basic pricing, pricing fluctuation range, and discount range parameters; S52. Define the objective function as: Among them, F is the objective function value, P i is the profit of the i-th pricing strategy, C i is the cost of the i-th pricing strategy, w i is the weight, ΔP j is the price fluctuation range of the j-th strategy, γ is the fluctuation penalty coefficient, L is the local search improvement term, κ is the local improvement weight, n is the number of strategies, and m is the number of price fluctuation items; S53, initialize the position and velocity of each particle in the subgroup. The initial position of each particle Randomly distributed in the predetermined search space, the initial speed satisfy: Among them, U(-V max ,V max ) is in the interval [-V max ,V max ] uniform distribution on ; S54, update the particle velocity, using the dynamic inertia weight w(k) and the local gradient correction term, the update formula is: in, is the velocity of the ith particle in the kth generation, is the position of the ith particle, is the best historical position of the particle, is the optimal position of the subgroup, c1, c2, c3 are learning factors, r1, r2 are random variables, At the particle position A gradient correction term based on local search; S55. Use a local search strategy for particles in the subgroup, use the gradient descent method to fine-tune the local optimal solution, calculate the local improvement term L so that the local search obtains a better pricing strategy, and adjust the current particle position: in, is the position of the ith particle in the kth generation, is the updated particle velocity, is the position of the i-th particle in the k+1th generation; S56. According to the search results of the particle swarm, the pricing strategy of the global optimal particle is adopted as the current optimization strategy. The update formula of the global optimal solution is: in, is the position of the global optimal particle, is the fitness value of the i-th particle in the k+1th generation; S57. Dynamically adjust the size and search range of the particle swarm during the optimization process, and make adaptive adjustments based on market feedback and optimization results. The particle swarm size and search range adjustment formula is: Among them, N new is the new particle swarm size, N old is the current particle swarm size, β is the adjustment coefficient, ΔF i is the fitness change of the ith particle, Represents the sum of the changes in fitness of all particles; S58, when the fitness value of the global optimal particle When the set threshold θ is reached, the optimization process is terminated and the optimal pricing strategy is output. The judgment formula of the termination condition is: Among them, Terminate is the termination condition flag, True means triggering the optimization process to stop, and False means not triggering the optimization process to stop.

7. The method for dynamic pricing of online charging data fusion platform based on adaptive optimization according to claim 1 is characterized in that: The S6 comprises the following steps: S61. Automatically implement dynamic pricing based on the optimized pricing strategy. The pricing adjustment formula is: New_Price=Base_Price+ΔP+ΔU+ΔT; Among them, New_Price is the adjusted price, Base_Price is the basic price, ΔP is the price fluctuation adjustment generated according to market feedback, ΔU is the personalized price adjustment generated according to user consumption behavior, and ΔT is the time period adjustment generated according to the time period; S62. Calculate the personalized pricing adjustment value for each user using a weighted average method based on the user's historical behavior data; S63. Adjust the time period based on user behavior and market feedback. The time period adjustment formula is: Among them, ΔT is the time period adjustment, α and β are adjustment coefficients, and Demand is the demand; S64. Apply the dynamic pricing generated by combining various adjustment values ​​to the charging platform.

8. The method for dynamic pricing of online charging data fusion platform based on adaptive optimization according to claim 1 is characterized in that: The S7 comprises the following steps: S71. Real-time monitoring of pricing implementation effects, the monitoring content includes platform revenue, user satisfaction, transaction volume, price fluctuation indicators, and the monitoring formula is: Among them, M is the overall monitoring index, m i is the i-th monitoring indicator, w i is the weight of the i-th monitoring indicator, and n is the number of monitoring indicators; S72. Process the collected feedback data and calculate the change rate of each monitoring indicator. The change rate formula is: Among them, Δm i The rate of change of the i-th monitoring indicator, is the new period indicator value, is the old period indicator value; S73. Determine whether the current pricing strategy has achieved the expected goal based on the change rate of the monitoring data. If the change rate exceeds the set threshold, trigger the adjustment mechanism. S74. Set a feedback threshold. If the user satisfaction in the feedback data is lower than the feedback threshold, adjust the pricing strategy according to market demand fluctuations and user behavior. The adjustment formula is: ΔP=α·(Feedback-Threshold); Among them, ΔP is the adjusted pricing change, α is the adjustment coefficient, Feedback is the user satisfaction, and Threshold is the preset feedback threshold; S75. Update the parameters of the optimization model in real time based on market and user feedback.

Images